图像分割工作总结（五）opencv遍历图像

来源：互联网发布：sql语句中union 编辑：程序博客网时间：2024/05/01 17:01

(1)间接访问（调用函数访问）效率不高

IplImage*src=cvLoadImage(filename,0);CvScalar s=cvGet2D(src,i,j);printf("%f",s.val[0]);cvSet2D(src,i,j,s);

（2）直接访问(单通道单字节)

IplImage* src=cvLoadImage(filename,0);(uchar*)(src->imagedata+h*src->widthstep)[w]=111;

（3）指针访问(单通道)

uchar* data=(uchar*)src->imagedata;int step=src->widthstep/sizeof(uchar);data[h*step+w]=111;

指针访问（多通道）

float*data=(float*)src->imagedata;int step=src->widthstep/sizeof(float);data[h*step+w*channels+k]=111;

(4)利用C++模板

template<class T> class Image{    public:          IplImage* pimg;          Image(IplImage* img=0){pimg   =  img;}          ~Image(){pimg=0;}           void operater=(pimg=img;)           inline T* operater[](const int row)         {return ((T*)(pimg->imagedata+row*pimg->widthstep));}}struct RgbPixel{  unsigned char r,g,b;};struct RgbPixelFloat{   float r,g,b;}Image<unsigned char> imageA（src）;imageA[i][j]=111;

0 0